Under the Smart City Project, Rourkela (the third-largest city in Odisha after Bhubaneswar and Cuttack) is all set to come under facial recognition surveillance and artificial intelligence-based video analytics for ‘tracking people’, ‘recognising patterns of demonstration in crowds’, and so on.
The facial recognition system will be integrated with IP video cameras (CCTVs), and it should be able to recognise subjects appearing simultaneously in multiple live video streams, the Odisha government said. “The system should also have the ability to handle initial real-time watchlist of 10,000 faces (should be scalable to at least 1 million faces) and 50 camera feeds simultaneously and generate face matching alerts,” it added.
These details are part of a tender floated by Rourkela Smart City Limited, a special purpose vehicle of the Government of Odisha for the selection of a master system integrator for ‘Implementation of Smart Solutions in Rourkela city’. Apart from that, the facial recognition system will also be used to assign ‘different security levels to people and places’. “It should alert security staff when someone is spotted in an area where they’re not permitted, whilst allowing them free access to nonrestricted/ public areas,” the tender read.
Overall, these applications of artificial intelligence and facial recognition paint a grim picture of the present state of surveillance in India and where it’s headed in the absence of robust data privacy laws. However, even if the current form of the Personal Data Protection Bill is adopted as a law in the future, its various clauses say that government agencies, the Odisha government in this case, would be exempted from its provisions.
System should be able to detect faces regardless of vantage points
The tender said that the algorithm for facial recognition should be able to recognise partial faces at varying angles. It should also be able to detect multiple faces from a live single video feed. “The system shall be able to detect faces in different environmental changes like rain, wind, fog and poor light,” it added.
“Face detection algorithms, modes and search depths should be suitable for different environments such as fast detection, high accuracy etc. The FRS system shall make use of advance technology for high performance and computing with artificial intelligence with continuous improvement instead of traditional CPUs and Computer vision, to greatly improve the computational performance in crowded environments,” the tender read.
Apart from that, the tender required that the facial recognition system should —
- Identify and authenticate based on individual facial features
- Be compatible with the video management system being proposed by the system integrator
- Have capability for 1:1 verification and 1:N identification matching
- Integrate with other systems in the future such as ‘Automatic fingerprint identification system (AFIS)’ etc.
- Support diverse industry-standard graphic and video formats as well as live cameras
- Match faces from recorded media
- Detect a face from a group photo
- Detect a face from stored videos of any format
- Have a bulk process of adding faces in the system
- Be an independent system, with the capability to integrate with industry-standard Video Management Systems (VMS) for alert viewing
Video analytics to be installed in 110 cameras across the city
The tender described a few scenarios wherein artificial intelligence-based video analytics would come in handy. “These use cases are to be implemented using AI through various cameras, sensors implemented in the field..with continuous learning capabilities at Data Centre with 85% accuracy in all below cases,” the tender said. The use cases mentioned in the tender are:
- Graffiti and vandalism detection, defacing/destroying the public property, and street furniture.
- Detection and recognition of the pattern of demonstration and conflicts in crowds
- Detection and classification of human, animal, and vehicle
- Parking violation
- Speeding violation
- Accident detection
- Loitering detection
- Person climbing barricade
- Person collapsing
- Gesture recognition: Identification through gesture change
- Vehicle of interest’ tracking by colour, speed, number plate
- Helmet detection on two-wheelers
- Unwanted/ banned vehicle detection
- Wrong-way or illegal turn detection
- Public toilet cleaning and the number of people using public toilets
- Environmental condition detection
- Fire detection at minimum 50 plus high-risk locations.
- Monitoring of underbridge storm waterlogging.
- People loitering
- Unattended object
- People tracking
What about the privacy of citizens?
“The architecture must adopt an end-to-end security model that protects data and the infrastructure from malicious attacks, theft, natural disasters etc. MSI must make provisions for security of field equipment as well as protection of the software system from hackers and other threats,” the tender read.
The Odisha government, in the tender, recommended that firewalls and intrusion prevention systems should be installed to prevent attacks. “The virus and worm attacks should be well defended with gateway level antivirus system, along with workstation level anti-virus mechanism,” the tender read.
Apart from that, the tender also recommended that the system integrator use VPN technologies to secure communications between applications and their end-users. “Furthermore, all the system logs should be properly stored & archived for future analysis and forensics whenever desired. The authority would carry out the security audit of the entire system upon handover and also at regular intervals during O&M period,” it added.
Apart from that, the tender said —
- 3 factors (User ID & Password, Biometric, and Digital Signature) security mechanisms should be implemented to enable secure login and authorised access to portal information and services.
- Role-based access for all the stakeholders to be implemented to access and use the system
India needs a well-planned regulatory framework for bulk targeting data surveillance
Where do these facial recognition/video analytics surveillance projects stand in the absence of a data protection law in the country? “The bigger question is how does one mitigate the high risks that are associated with the development and procurement of such surveillance technologies. The question today is really not the lack of personal data protection law in the country. We see that despite having developed data protection laws such as in the UK, there have been instances of misuse irrespective of such laws being in place, such as affecting communities of colour under circumstances and political regimes,” Anushkaa Arora, Principal & Founder, ABA Law Office said.
Then, how does one mitigate privacy risks? By developing a regulatory framework on the principle of proportionality in terms of the apex court’s Puttaswamy judgement (2017), oversight and accountability, rights of the persons who are under surveillance, etc. In addition to this, deployment of surveillance tools would mean the deployment of CCTVs, so devices in bulk; hence, a comprehensive and harmonised regulation of CCTVs is needed before such tools are deployed, Arora said.
Counterpoint: “Surveillance is an important tool for maintaining the security of the state and helps in investigation and prevention of any mishappenings,” Amay Jain, Senior Associate, Victoriam Legalis – Advocates & Solicitors said.
But data being collected is not protected against misuse: “There is no information as to who all can access such data and whether or not it is being used as per the decided objectives. There are no measures to tackle or prevent such misuse. As far as mass surveillance is concerned, it clearly violates the right of privacy guaranteed under the Constitution of India. Technology and digital surveillance are important for the protection of citizens but we need a proper regulatory framework in place, not just for the protection of privacy but for surveillance too. Data Protection Bill 2021 allows more on controlling the weak technology practices rather than protecting the privacy aspects,” he added.
What will be the future of surveillance in India?
Do you want to keep track of facial recognition usage in India but don’t have the time? Relying on scattered content from across the web makes it feel harder than it needs to be.
Subscribe to MediaNama and get crisp, timely updates on tech policy developments in India and across the world.
- Exclusive: Major entrance examinations such as JEE, NEET, UGC-NET to come under facial recognition surveillance
- 1 lakh CCTVs at 4,000 centers with facial recognition: National Testing Agency expands surveillance of JEE, NEET exams
- MoHFW proposes facial recognition for verification of candidates sitting for exams conducted by AIIMS
Have something to add? Subscribe to MediaNama here and post your comment.